Rough set approach to predict biochar stability and pH from pyrolysis conditions and feedstock characteristics.

Biochar has gained increasing attention as a potential adsorbent for the sequestration of carbon dioxide (CO 2) and enhancement of soil fertility. Biochar pH and stability are important adsorbent properties as they indicate the affinity of biochar for CO 2 and its potential to be applied in soil for long periods. These attributes are influenced by the feedstock composition and pyrolysis conditions. Therefore, it is important to develop a model that can elucidate the underlying trends and inherent relationships between feedstock composition and pyrolysis operating parameters on biochar pH and stability. In this work, rough set machine learning (RSML) tools have been used to develop a model to quantify this relationship because of the interpretable nature of RSML. RSML is a rule-based prediction model that categorizes the biochar properties by utilizing 'if. then' rules to conditional attributes. In this study, the feedstock properties such as elemental (carbon, hydrogen, oxygen and nitrogen) composition, fixed carbon, ash content, volatile matter, and operating conditions such as residence time, heating rate, and temperature were the conditional attributes while pH, ash content and O/C ratio of biochar were considered the decision attributes. The rules generated from RSML were validated and evaluated for scientific coherency. Thus, this approach provided a model which could reflect on the physical phenomena at varying process conditions. As a result, it was recommended that the pyrolysis temperature is in between 375 and 475 °C, the ash content in the feedstock is between 2.59 and 3.55 wt% and volatile matter in the feedstock is in between 68.9 and 73.8 wt% to obtain biochar with minimum ash content (0–5 wt%), minimum O/C ratio (0–0.2), and high pH (9−11). [Display omitted] • Rough-set ML to predict biochar pH and stability from pyrolysis and feedstock. • Rules with certainties >60% and ensured its scientific coherency. • Higher T for higher pH and O/C ratios; low T for biochar ash content. • Higher AC leads to higher O/C ratios and biochar ash content. • Low VM leads high biochar pH and low ash content. [ABSTRACT FROM AUTHOR]